Guest_Jim_* Posted June 1, 2019 Posted June 1, 2019 16 minutes ago, Braegnok said: Looks like the frequencies in the gaming loop are about 100 MHz higher on the gaming OC card than the reference card. The specs page says it is more on the order of 20-30 MHz, but regardless I will be happy with the better and potentially quieter cooler. Share this post Link to post Share on other sites More sharing options...
Braegnok Posted June 1, 2019 Posted June 1, 2019 (edited) My bad,.. I forgot to include the review link. Tom's does a Temperatures vs. Frequency comparison,.. during GPU review in both Gaming Loop & Torture Loop. The results in both tests look similar. Edited September 27, 2020 by Braegnok Share this post Link to post Share on other sites More sharing options...
Guest_Jim_* Posted June 11, 2019 Posted June 11, 2019 The replacement arrived today so I will be sticking it in the test system to make sure everything is good with it (and maybe see about undervolting) later. Eventually I may stick it back in the desktop, but not just yet as I don't feel like taking the desktop down and apart at the moment. Share this post Link to post Share on other sites More sharing options...
Guest_Jim_* Posted June 11, 2019 Posted June 11, 2019 And this is definitely one advantage to having the test system. Been having some stability issues with the replacement, so I'm going to try reaching out to Gigabyte about it. Initially it was getting through 3DMark stress tests, but then I started experimenting with WattMan like setting a custom fan curve and raising the power limit, and then the tests would crash the system. Now even after a clean install of the drivers, it crashes immediately upon starting the tests. Share this post Link to post Share on other sites More sharing options...
ir_cow Posted June 11, 2019 Posted June 11, 2019 Do you have a spare power supply? Voltage drop is real thing! A faulty power supply can still operate even under light load sometimes. Before chalking it up to another bad card, try a different psu if you have one. Share this post Link to post Share on other sites More sharing options...
Guest_Jim_* Posted June 12, 2019 Posted June 12, 2019 Don't have a spare PSU around, but as it has been able to handle the original Vega and an RTX 2080, both under the load of 3DMark stress tests (and I ran the RTX 2080 through a test immediately after I took out the replacement Vega to ensure the system was fine), I am inclined to doubt the issue is with the system's other components. I could definitely be wrong, but without doing deeper analysis than I'm able to currently, that's the best I have for now. Gigabyte did already get back to me though, with a new RMA number and shipping label, so it will be headed back. Share this post Link to post Share on other sites More sharing options...
ir_cow Posted June 13, 2019 Posted June 13, 2019 I think your right. If a RTX 2080 can stay stable under load, than it is unlikely it would be anything else in the system besides the card. You just got bad luck I guess :/ Share this post Link to post Share on other sites More sharing options...
Guest_Jim_* Posted June 13, 2019 Posted June 13, 2019 11 hours ago, ir_cow said: I think your right. If a RTX 2080 can stay stable under load, than it is unlikely it would be anything else in the system besides the card. You just got bad luck I guess :/ I know I have bad luck, but at least for this example it could be worse. I mean, I have a selection of graphics cards I can go to if I need to, to keep up and running. I also haven't the need to collect data at the moment either, so lacking a Vega 64 isn't too big an issue there either, but hopefully this can all be resolved soon. Share this post Link to post Share on other sites More sharing options...
Guest_Jim_* Posted July 2, 2019 Posted July 2, 2019 Have been testing the new replacement, and hardware-wise seems very happy with everything I have been doing to it. Only issue is the HBM2 does not want to stay at its top speed (945 MHz at stock, and I'm not overclocking, just undervolting) but that has an easy fix in Wattman. You can set P-states to be the minimum or maximum, so with that top state being set to the minimum, it cannot be slower at any time. Not really ideal, but it works and even if it technically may increase power usage and temperature at idle, if I just leave it in the test system instead of the desktop, that will not be a problem. Going to keep testing, but by now the previous replacement was already clearly busted. Share this post Link to post Share on other sites More sharing options...
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now